This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Upgrade to Module::CoreList 2.09:
[perl5.git] / regcomp.h
CommitLineData
a0d0e21e 1/* regcomp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
1d325971 4 * 2000, 2001, 2002, 2003, 2005 by Larry Wall and others
d6376244
JH
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
a687059c
LW
9 */
10
c277df42
IZ
11typedef OP OP_4tree; /* Will be redefined later. */
12
07be1b83
YO
13
14#define PERL_ENABLE_TRIE_OPTIMISATION 1
15#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
1de06328 16#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 1
07be1b83
YO
17#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
18
a687059c
LW
19/*
20 * The "internal use only" fields in regexp.h are present to pass info from
21 * compile to execute that permits the execute phase to run lots faster on
22 * simple cases. They are:
23 *
bd61b366 24 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
25 * reganch is the match anchored (at beginning-of-line only)?
26 * regmust string (pointer into program) that match must include, or NULL
79072805 27 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
28 * regmlen length of regmust string
29 * [regmlen not used currently]
30 *
31 * Regstart and reganch permit very fast decisions on suitable starting points
32 * for a match, cutting down the work a lot. Regmust permits fast rejection
33 * of lines that cannot possibly match. The regmust tests are costly enough
e50aee73 34 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
35 * potentially expensive (at present, the only such thing detected is * or +
36 * at the start of the r.e., which can involve a lot of backup). Regmlen is
e50aee73 37 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
38 * it anyway.
39 * [regmust is now supplied always. The tests that use regmust have a
40 * heuristic that disables the test if it usually matches.]
41 *
42 * [In fact, we now use regmust in many cases to locate where the search
43 * starts in the string, so if regback is >= 0, the regmust search is never
44 * wasted effort. The regback variable says how many characters back from
45 * where regmust matched is the earliest possible start of the match.
46 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
47 */
48
49/*
50 * Structure for regexp "program". This is essentially a linear encoding
51 * of a nondeterministic finite-state machine (aka syntax charts or
52 * "railroad normal form" in parsing technology). Each node is an opcode
53 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
54 * all nodes except BRANCH implement concatenation; a "next" pointer with
55 * a BRANCH on both ends of it is connecting two alternatives. (Here we
56 * have one of the subtle syntax dependencies: an individual BRANCH (as
57 * opposed to a collection of them) is never concatenated with anything
58 * because of operator precedence.) The operand of some types of node is
59 * a literal string; for others, it is a node leading into a sub-FSM. In
60 * particular, the operand of a BRANCH node is the first node of the branch.
61 * (NB this is *not* a tree structure: the tail of the branch connects
a3621e74
YO
62 * to the thing following the set of BRANCHes.) The opcodes are defined
63 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
a687059c
LW
64 */
65
a687059c
LW
66/*
67 * A node is one char of opcode followed by two chars of "next" pointer.
68 * "Next" pointers are stored as two 8-bit pieces, high order first. The
69 * value is a positive offset from the opcode of the node containing it.
70 * An operand, if any, simply follows the node. (Note that much of the
71 * code generation knows about this implicit relationship.)
72 *
73 * Using two bytes for the "next" pointer is vast overkill for most things,
74 * but allows patterns to get big without disasters.
75 *
e91177ed 76 * [The "next" pointer is always aligned on an even
a687059c
LW
77 * boundary, and reads the offset directly as a short. Also, there is no
78 * special test to reverse the sign of BACK pointers since the offset is
79 * stored negative.]
80 */
81
c277df42 82struct regnode_string {
cd439c50 83 U8 str_len;
c277df42
IZ
84 U8 type;
85 U16 next_off;
cd439c50 86 char string[1];
c277df42 87};
a687059c 88
c277df42
IZ
89struct regnode_1 {
90 U8 flags;
91 U8 type;
92 U16 next_off;
93 U32 arg1;
94};
95
96struct regnode_2 {
97 U8 flags;
98 U8 type;
99 U16 next_off;
100 U16 arg1;
101 U16 arg2;
102};
103
936ed897 104#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
ffc61ed2 105#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 106
786e8c11 107/* also used by trie */
936ed897
IZ
108struct regnode_charclass {
109 U8 flags;
110 U8 type;
111 U16 next_off;
ffc61ed2 112 U32 arg1;
19103b91 113 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
114};
115
19103b91
JH
116struct regnode_charclass_class { /* has [[:blah:]] classes */
117 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
118 U8 type;
119 U16 next_off;
ffc61ed2 120 U32 arg1;
19103b91
JH
121 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
122 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
123};
124
9c7e81e8
AD
125/* XXX fix this description.
126 Impose a limit of REG_INFTY on various pattern matching operations
127 to limit stack growth and to avoid "infinite" recursions.
128*/
129/* The default size for REG_INFTY is I16_MAX, which is the same as
130 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
131 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
132 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
133 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
134 smaller default.
135 --Andy Dougherty 11 June 1998
136*/
137#if SHORTSIZE > 2
138# ifndef REG_INFTY
139# define REG_INFTY ((1<<15)-1)
140# endif
141#endif
142
143#ifndef REG_INFTY
83cfe48c
IZ
144# define REG_INFTY I16_MAX
145#endif
a687059c 146
e91177ed
IZ
147#define ARG_VALUE(arg) (arg)
148#define ARG__SET(arg,val) ((arg) = (val))
c277df42 149
e8e3b0fb
JH
150#undef ARG
151#undef ARG1
152#undef ARG2
153
c277df42
IZ
154#define ARG(p) ARG_VALUE(ARG_LOC(p))
155#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
156#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
786e8c11 157
c277df42
IZ
158#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
159#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
160#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
161
e8e3b0fb
JH
162#undef NEXT_OFF
163#undef NODE_ALIGN
164
075abff3
AL
165#define NEXT_OFF(p) ((p)->next_off)
166#define NODE_ALIGN(node)
167#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 168
c277df42
IZ
169#define SIZE_ALIGN NODE_ALIGN
170
e8e3b0fb
JH
171#undef OP
172#undef OPERAND
173#undef MASK
174#undef STRING
175
e91177ed
IZ
176#define OP(p) ((p)->type)
177#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
178#define MASK(p) ((char*)OPERAND(p))
179#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
180#define STRING(p) (((struct regnode_string *)p)->string)
181#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
182#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
183
e8e3b0fb
JH
184#undef NODE_ALIGN
185#undef ARG_LOC
186#undef NEXTOPER
187#undef PREVOPER
188
e91177ed
IZ
189#define NODE_ALIGN(node)
190#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
191#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
192#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
786e8c11
YO
193
194
e91177ed
IZ
195#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
196#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
197
198#define NODE_STEP_B 4
c277df42
IZ
199
200#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
201#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
202
e91177ed 203#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 204 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 205#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 206 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 207
22c35a8c 208#define REG_MAGIC 0234
a687059c 209
830247a4 210#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 211
936ed897 212/* Flags for node->flags of ANYOF */
b8c5462f 213
19103b91 214#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
215#define ANYOF_INVERT 0x04
216#define ANYOF_FOLD 0x02
217#define ANYOF_LOCALE 0x01
b8c5462f 218
653099ff 219/* Used for regstclass only */
ffc61ed2
JH
220#define ANYOF_EOS 0x10 /* Can match an empty string too */
221
222/* There is a character or a range past 0xff */
223#define ANYOF_UNICODE 0x20
1aa99e6b 224#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 225
4f66b38d
HS
226/* size of node is large (includes class pointer) */
227#define ANYOF_LARGE 0x80
228
ffc61ed2
JH
229/* Are there any runtime flags on in this node? */
230#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
231
232#define ANYOF_FLAGS_ALL 0xff
653099ff 233
936ed897 234/* Character classes for node->classflags of ANYOF */
653099ff
GS
235/* Should be synchronized with a table in regprop() */
236/* 2n should pair with 2n+1 */
b8c5462f 237
289d4f09 238#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 239#define ANYOF_NALNUM 1
aaa51d5e 240#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
241#define ANYOF_NSPACE 3
242#define ANYOF_DIGIT 4
243#define ANYOF_NDIGIT 5
289d4f09 244#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
245#define ANYOF_NALNUMC 7
246#define ANYOF_ALPHA 8
247#define ANYOF_NALPHA 9
248#define ANYOF_ASCII 10
249#define ANYOF_NASCII 11
250#define ANYOF_CNTRL 12
251#define ANYOF_NCNTRL 13
252#define ANYOF_GRAPH 14
253#define ANYOF_NGRAPH 15
254#define ANYOF_LOWER 16
255#define ANYOF_NLOWER 17
256#define ANYOF_PRINT 18
257#define ANYOF_NPRINT 19
258#define ANYOF_PUNCT 20
259#define ANYOF_NPUNCT 21
260#define ANYOF_UPPER 22
261#define ANYOF_NUPPER 23
262#define ANYOF_XDIGIT 24
263#define ANYOF_NXDIGIT 25
aaa51d5e
JF
264#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
265#define ANYOF_NPSXSPC 27
ffc61ed2 266#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 267#define ANYOF_NBLANK 29
b8c5462f 268
aaa51d5e 269#define ANYOF_MAX 32
b8c5462f
JH
270
271/* Backward source code compatibility. */
272
273#define ANYOF_ALNUML ANYOF_ALNUM
274#define ANYOF_NALNUML ANYOF_NALNUM
275#define ANYOF_SPACEL ANYOF_SPACE
276#define ANYOF_NSPACEL ANYOF_NSPACE
277
278/* Utility macros for the bitmap and classes of ANYOF */
279
936ed897
IZ
280#define ANYOF_SIZE (sizeof(struct regnode_charclass))
281#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 282
936ed897 283#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
284
285#define ANYOF_BIT(c) (1 << ((c) & 7))
286
936ed897 287#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
288#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
289#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
290#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
291
936ed897
IZ
292#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
293#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
294
295#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 296#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
297#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
298#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
299#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
300
f8bef550
NC
301#define ANYOF_BITMAP_SETALL(p) \
302 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
303#define ANYOF_BITMAP_CLEARALL(p) \
304 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
305/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
306#define ANYOF_BITMAP_TESTALLSET(p) \
307 memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE)
308
936ed897
IZ
309#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
310#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
311#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 312
786e8c11 313
a687059c
LW
314/*
315 * Utility definitions.
316 */
2304df62 317#ifndef CHARMASK
075abff3 318# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 319#else
075abff3 320# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 321#endif
a687059c 322
c277df42
IZ
323#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
324
3dab1dad
YO
325#define REG_SEEN_ZERO_LEN 0x00000001
326#define REG_SEEN_LOOKBEHIND 0x00000002
327#define REG_SEEN_GPOS 0x00000004
328#define REG_SEEN_EVAL 0x00000008
329#define REG_SEEN_CANY 0x00000010
cce850e4 330#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
d09b2d29 331
73c4f7a1
GS
332START_EXTERN_C
333
be8e71aa
YO
334#ifdef PLUGGABLE_RE_EXTENSION
335#include "re_nodes.h"
336#else
d09b2d29 337#include "regnodes.h"
be8e71aa 338#endif
d09b2d29 339
4d61ec05 340/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 341#ifndef DOINIT
4d61ec05 342EXTCONST U8 PL_varies[];
d09b2d29 343#else
4d61ec05 344EXTCONST U8 PL_varies[] = {
0111c4fd 345 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
a0ed51b3 346 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0
c277df42 347};
d09b2d29 348#endif
c277df42 349
4d61ec05
GS
350/* The following always have a length of 1. U8 we can do strchr() on it. */
351/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 352#ifndef DOINIT
4d61ec05 353EXTCONST U8 PL_simple[];
d09b2d29 354#else
4d61ec05 355EXTCONST U8 PL_simple[] = {
f33976b4 356 REG_ANY, SANY, CANY,
ffc61ed2
JH
357 ANYOF,
358 ALNUM, ALNUML,
359 NALNUM, NALNUML,
360 SPACE, SPACEL,
361 NSPACE, NSPACEL,
362 DIGIT, NDIGIT,
363 0
c277df42
IZ
364};
365#endif
366
f9f4320a
YO
367#ifndef PLUGGABLE_RE_EXTENSION
368#ifndef DOINIT
369EXTCONST regexp_engine PL_core_reg_engine;
370#else /* DOINIT */
371EXTCONST regexp_engine PL_core_reg_engine = {
372 Perl_pregcomp,
373 Perl_regexec_flags,
374 Perl_re_intuit_start,
375 Perl_re_intuit_string,
376 Perl_pregfree,
377#if defined(USE_ITHREADS)
378 Perl_regdupe
379#endif
380};
381#endif /* DOINIT */
382#endif /* PLUGGABLE_RE_EXTENSION */
383
384
73c4f7a1 385END_EXTERN_C
cad2e5aa 386
cad2e5aa 387
261faec3
MJD
388/* .what is a character array with one character for each member of .data
389 * The character describes the function of the corresponding .data item:
0111c4fd 390 * f - start-class data for regstclass optimization
261faec3
MJD
391 * n - Root of op tree for (?{EVAL}) item
392 * o - Start op for (?{EVAL}) item
3dab1dad 393 * p - Pad for (?{EVAL}) item
9e55ce06
JH
394 * s - swash for unicode-style character class, and the multicharacter
395 * strings resulting from casefolding the single-character entries
396 * in the character class
a3621e74 397 * t - trie struct
07be1b83 398 * T - aho-trie struct
261faec3 399 * 20010712 mjd@plover.com
c3613ed7 400 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 401 */
cad2e5aa
JH
402struct reg_data {
403 U32 count;
404 U8 *what;
405 void* data[1];
406};
407
408struct reg_substr_datum {
409 I32 min_offset;
410 I32 max_offset;
33b8afdf
JH
411 SV *substr; /* non-utf8 variant */
412 SV *utf8_substr; /* utf8 variant */
1de06328 413 I32 end_shift;
cad2e5aa
JH
414};
415
416struct reg_substr_data {
417 struct reg_substr_datum data[3]; /* Actual array */
418};
419
420#define anchored_substr substrs->data[0].substr
33b8afdf 421#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 422#define anchored_offset substrs->data[0].min_offset
1de06328
YO
423#define anchored_end_shift substrs->data[0].end_shift
424
cad2e5aa 425#define float_substr substrs->data[1].substr
33b8afdf 426#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
427#define float_min_offset substrs->data[1].min_offset
428#define float_max_offset substrs->data[1].max_offset
1de06328
YO
429#define float_end_shift substrs->data[1].end_shift
430
cad2e5aa 431#define check_substr substrs->data[2].substr
33b8afdf 432#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
433#define check_offset_min substrs->data[2].min_offset
434#define check_offset_max substrs->data[2].max_offset
1de06328 435#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
436
437
438
439/* trie related stuff */
5d9a96ca 440
a3621e74
YO
441/* a transition record for the state machine. the
442 check field determines which state "owns" the
443 transition. the char the transition is for is
444 determined by offset from the owning states base
445 field. the next field determines which state
446 is to be transitioned to if any.
447*/
448struct _reg_trie_trans {
449 U32 next;
450 U32 check;
451};
452
453/* a transition list element for the list based representation */
454struct _reg_trie_trans_list_elem {
455 U16 forid;
456 U32 newstate;
457};
458typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
459
460/* a state for compressed nodes. base is an offset
461 into an array of reg_trie_trans array. If wordnum is
462 nonzero the state is accepting. if base is zero then
463 the state has no children (and will be accepting)
464*/
465struct _reg_trie_state {
466 U16 wordnum;
467 union {
468 U32 base;
469 reg_trie_trans_le* list;
470 } trans;
471};
472
473
474
a3621e74
YO
475typedef struct _reg_trie_state reg_trie_state;
476typedef struct _reg_trie_trans reg_trie_trans;
477
478
479/* anything in here that needs to be freed later
8e11feef 480 should be dealt with in pregfree */
a3621e74 481struct _reg_trie_data {
786e8c11
YO
482 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
483 U32 lasttrans; /* last valid transition element */
484 U16 *charmap; /* byte to charid lookup array */
485 HV *widecharmap; /* code points > 255 to charid */
486 reg_trie_state *states; /* state data */
487 reg_trie_trans *trans; /* array of transition elements */
488 char *bitmap; /* stclass bitmap */
489 U32 refcount; /* number of times this trie is referenced */
490 U32 startstate; /* initial state - used for common prefix optimisation */
491 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
492 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
493 U32 *wordlen; /* array of lengths of words */
494 U16 *jump; /* optional 1 indexed array of offsets before tail
495 for the node following a given word. */
496 U16 *nextword; /* optional 1 indexed array to support linked list
497 of duplicate wordnums */
498 U32 laststate; /* Build only */
499 U32 wordcount; /* Build only */
a3621e74 500#ifdef DEBUGGING
786e8c11
YO
501 STRLEN charcount; /* Build only */
502 AV *words; /* Array of words contained in trie, for dumping */
503 AV *revcharmap; /* Map of each charid back to its character representation */
a3621e74
YO
504#endif
505};
a3621e74
YO
506typedef struct _reg_trie_data reg_trie_data;
507
07be1b83
YO
508struct _reg_ac_data {
509 U32 *fail;
510 reg_trie_state *states;
511 reg_trie_data *trie;
512 U32 refcount;
513};
514typedef struct _reg_ac_data reg_ac_data;
515
3dab1dad
YO
516/* ANY_BIT doesnt use the structure, so we can borrow it here.
517 This is simpler than refactoring all of it as wed end up with
518 three different sets... */
519
520#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 521#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
522#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
523#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
524#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 525
1de06328
YO
526#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
527#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
528
529
786e8c11
YO
530#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
531#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 532
a3621e74
YO
533/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
534#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
535#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
536#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
537
3dab1dad 538#ifdef DEBUGGING
3dab1dad 539#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad
YO
540#define TRIE_REVCHARMAP(trie) ((trie)->revcharmap)
541#else
3dab1dad 542#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad
YO
543#define TRIE_REVCHARMAP(trie) (trie_revcharmap)
544#endif
a3621e74 545
0111c4fd
RGS
546#define RE_TRIE_MAXBUF_INIT 65536
547#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
548#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
549
be8e71aa
YO
550/*
551
552RE_DEBUG_FLAGS is used to control what debug output is emitted
553its divided into three groups of options, some of which interact.
554The three groups are: Compile, Execute, Extra. There is room for a
555further group, as currently only the low three bytes are used.
556
557 Compile Options:
07be1b83 558
be8e71aa
YO
559 PARSE
560 PEEP
561 TRIE
562 PROGRAM
563 OFFSETS
a3621e74 564
be8e71aa 565 Execute Options:
a3621e74 566
be8e71aa
YO
567 INTUIT
568 MATCH
569 TRIE
a3621e74 570
be8e71aa 571 Extra Options
a3621e74 572
be8e71aa
YO
573 TRIE
574 OFFSETS
575
576If you modify any of these make sure you make corresponding changes to
577re.pm, especially to the documentation.
a3621e74 578
be8e71aa
YO
579*/
580
581
582/* Compile */
583#define RE_DEBUG_COMPILE_MASK 0x0000FF
584#define RE_DEBUG_COMPILE_PARSE 0x000001
585#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
586#define RE_DEBUG_COMPILE_TRIE 0x000004
587#define RE_DEBUG_COMPILE_DUMP 0x000008
be8e71aa
YO
588
589/* Execute */
590#define RE_DEBUG_EXECUTE_MASK 0x00FF00
591#define RE_DEBUG_EXECUTE_INTUIT 0x000100
592#define RE_DEBUG_EXECUTE_MATCH 0x000200
593#define RE_DEBUG_EXECUTE_TRIE 0x000400
594
595/* Extra */
596#define RE_DEBUG_EXTRA_MASK 0xFF0000
597#define RE_DEBUG_EXTRA_TRIE 0x010000
598#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
599#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
600#define RE_DEBUG_EXTRA_STATE 0x080000
601#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
be8e71aa 602
e68ec53f 603#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
604/* Compile */
605#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 606 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 607#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 608 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 609#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 610 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 611#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 612 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 613#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 614 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 615#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 616 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
617
618/* Execute */
619#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 620 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 621#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 622 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 623#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 624 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 625#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 626 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
627
628/* Extra */
629#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 630 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
631#define DEBUG_OFFSETS_r(x) DEBUG_r( \
632 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
633#define DEBUG_STATE_r(x) DEBUG_r( \
634 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
a5ca303d
YO
635#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
636 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
637 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 638#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 639 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
640 Perl_warn_nocontext x )
641#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 642 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 643#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 644 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
645
646#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 647 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
648 | RE_DEBUG_EXECUTE_TRIE )) x )
649
650/* initialization */
651/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
652#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
653 SV * re_debug_flags_sv = NULL; \
654 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
655 if (re_debug_flags_sv) { \
8e9a8a48 656 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 657 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 658 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
659 }\
660})
a3621e74
YO
661
662#ifdef DEBUGGING
ab3bbdeb 663
8e9a8a48 664#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
665
666#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
667 const char * const rpv = \
ddc5bc0f 668 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
669 PL_colors[(c1)],PL_colors[(c2)], \
670 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 671 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
672
673#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
674 const char * const rpv = \
ddc5bc0f 675 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb
YO
676 PL_colors[(c1)],PL_colors[(c2)], \
677 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
678
679#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
680 const char * const rpv = \
ddc5bc0f 681 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
682 PL_colors[0], PL_colors[1], \
683 ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \
684 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
685 )
686
687#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
688#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
689
690#else /* if not DEBUGGING */
691
a3621e74 692#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
693#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
694#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
695#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
696#define RE_SV_DUMPLEN(ItEm)
697#define RE_SV_TAIL(ItEm)
698
699#endif /* DEBUG RELATED DEFINES */
a3621e74 700